[OSM-talk] Could we just pause any wikidata edits for a month or two?

Ryszard Mikke ryszard.mikke at gmail.com
Sun Oct 22 20:20:50 UTC 2017


On 15 October 2017 at 16:05, Tomas Straupis <tomasstraupis at gmail.com> wrote:

>   Lets take an example. History of this hillfort:
>   http://www.openstreetmap.org/node/1717783246/history
>
>   What happened here:
> 1. I've added a hillfort object "Žagarės piliakalnis" (Žagarės hillfort).
> 2. Med fixed wikipedia tag (removed underscores - good change, my
> mistake fixed).
> 3. I've updated local Lithuanian heritage id's (fine, we use that for
> synchronisation).
> 4. rmikke added a wikidata entry... chrm... could be fine as we do not
> use wikidata at all, somebody else might, but... then
> 5. kartonage "corrected" wikidata tag (whatever, don't care about
> that) but also wikipedia tag! And let's look what was the change:
> "Žagarės piliakalnis" changed to "Žagarės antrasis piliakalnis". In
> English „Žagarės hillfort“ to „Žagarės SECOND hillfort". Reason stated
> for change is "pointing to disambiquity page" (so fixing wiki* tags
> according to Youris idea/tool). What is wrong here? Name tag is still
> "Žagarės I piliakalnis“ - Žagarės FIRST hillfort. And that is correct,
> because there is a second hillfort nearby called "Žagarės SECOND"
> hillfort:
>  http://www.openstreetmap.org/#map=19/56.35690/23.23084
>   But wikipedia link on this FIRST hillfort now points to the SECOND
> hillfort.
>
>   What happens then. I fixed wikipedia tag and removed wikidata entry
> so that somebody else would not come and "fix" it again (at least
> until we fix the actual problem). But rmikke runs his script again and
> adds the same wikidata value again. So this item will once again show
> up in Youri's tool and somebody else can try to "fix" it.
>
>   The actual "on the ground" problem here is an error in official
> heritage data. Heritage codes or names are mixed/swaped in several
> official sources. It is impossible to fix this stuff by simply looking
> at the OSM data (and even by simply looking at heritage data because
> it contradicts with data of archaeologists). We (local community) will
> have to contact heritage guys, archaeologists to find out who is
> wrong. So it cannot be fixed by somebody without local knowledge and
> without local contacts. And the problem is that Youri's tool gives a
> false impression that it CAN be fixed.
>
>   And this hillfort does show up in our local wikipedia error list
> (which is produced without any use of wikidata whatsoever) and is just
> waiting in queue to be fixed.
>
>   The points I'm showing here:
>   1. Error identification can be done without wikidata tags (and we
> already identify more errors like: no object in OSM or no coordinates
> in wikipedia, multiple objects in OSM for the same wikipedia page).
>   2. Error can not be fixed without local knowledge.
>   3. If it was only wikidata tag I would not have noticed the bad
> change, because there is no way I would somehow know what those 324657
> 897984 65465465 id's stand for. It is only because of wikipedia tag
> that I spotted the problem.
>
>   There was a very logical and practical advice somewhere in this
> thread. If you got approval in OSM-RU, why can't you do your
> experimenting/fixing there first? To the very end. When all (or almost
> all) wiki errors would be fixed in Russia, you could create a report
> about your work. It could then by compared to other processes of
> fixing wiki data and it would be possible for a specific community to
> choose the most appropriate method. And there would be less
> "toxicity", because you would not be forcing your way on people who
> are successfully doing it in a different way.
>
>
So, to sum up:
1) There was a link to disambiguation page that no one has corrected until
it was detected by Yuri's tool.
2) User kartonage has wrongly linked "Žagarės I piliakalnis" to "Žagarės II
piliakalnis" in Wikipedia.
3) You have reverted it back to disambiguation link and no wikidata=* tag
even though there is an established ground truth in the form of big
information tables in front of each of those hillforts with names "Žagarės
piliakalnis I" and "Žagarės piliakalnis II" in big letters.

Yet you think that wikidata=* tag is the problem here?



-- 
--
http://tnij.com/WyszukiwarkaRowerowa     http://jolanta.korwin-mikke.pl/
R.Mikke at pl.vwfsag.de        Ryszard.Mikke at gmail.com

دراجة أكبر
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20171022/a0b5bb43/attachment-0001.html>


More information about the talk mailing list